Engagement of Sirpa Inhibits Growth and Induces Programmed Cell Death in Acute Myeloid Leukemia Cells

Engagement of SIRPa Inhibits Growth and Induces Programmed Cell Death in Acute Myeloid Leukemia Cells

Mahban Irandoust1,2., Julian Alvarez Zarate3., Isabelle Hubeek1,2, Ellen M. van Beek3, Karin Schornagel3, Aart J. F. Broekhuizen1, Mercan Akyuz1, Arjan A. van de Loosdrecht2, Ruud Delwel4, Peter J. Valk4, Edwin Sonneveld5, Pamela Kearns6, Ursula Creutzig7, Dirk Reinhardt7, Eveline S. J. M. de Bont8, Eva A. Coenen9, Marry M. van den Heuvel-Eibrink9, C. Michel Zwaan9, Gertjan J. L. Kaspers1, Jacqueline Cloos1,2", Timo K. van den Berg3*" 1 Department of Pediatric Hematology/Oncology, VU University Medical Centre, Amsterdam, The Netherlands, 2 Department of Hematology, VU University Medical Centre, Amsterdam, The Netherlands, 3 Sanquin Research & Landsteiner Laboratory, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands, 4 Department of Hematology, Erasmus University Medical Center, Rotterdam, The Netherlands, 5 Dutch Childhood Oncology Group (DCOG), The Hague, The Netherlands, 6 School of Cancer Sciences, University of Birmingham, Birmingham, United Kingdom, 7 Department of Pediatric Hematology/Oncology, Medical School Hannover, Hannover, Germany, 8 University Medical Center Groningen, Groningen, The Netherlands, 9 Department of Pediatric Hematology/Oncology, Erasmus MC/Sophia Children’s Hospital, Rotterdam, The Netherlands

Abstract

Background: Recent studies show the importance of interactions between CD47 expressed on acute myeloid leukemia (AML) cells and the inhibitory immunoreceptor, signal regulatory protein-alpha (SIRPa) on macrophages. Although AML cells express SIRPa, its function has not been investigated in these cells. In this study we aimed to determine the role of the SIRPa in acute myeloid leukemia.

Design and Methods: We analyzed the expression of SIRPa, both on mRNA and protein level in AML patients and we further investigated whether the expression of SIRPa on two low SIRPa expressing AML cell lines could be upregulated upon differentiation of the cells. We determined the effect of chimeric SIRPa expression on tumor cell growth and programmed cell death by its triggering with an agonistic antibody in these cells. Moreover, we examined the efficacy of agonistic antibody in combination with established antileukemic drugs.

Results: By microarray analysis of an extensive cohort of primary AML samples, we demonstrated that SIRPa is differentially expressed in AML subgroups and its expression level is dependent on differentiation stage, with high levels in FAB M4/M5 AML and low levels in FAB M0–M3. Interestingly, AML patients with high SIRPa expression had a poor prognosis. Our results also showed that SIRPa is upregulated upon differentiation of NB4 and Kasumi cells. In addition, triggering of SIRPa with an agonistic antibody in the cells stably expressing chimeric SIRPa, led to inhibition of growth and induction of programmed cell death. Finally, the SIRPa-derived signaling synergized with the activity of established antileukemic drugs.

Conclusions: Our data indicate that triggering of SIRPa has antileukemic effect and may function as a potential therapeutic target in AML.

Citation: Irandoust M, Alvarez Zarate J, Hubeek I, van Beek EM, Schornagel K, et al. (2013) Engagement of SIRPa Inhibits Growth and Induces Programmed Cell Death in Acute Myeloid Leukemia Cells. PLoS ONE 8(1): e52143. doi:10.1371/journal.pone.0052143 Editor: David D. Roberts, Center for Cancer Research, National Cancer Institute, United States of America Received July 31, 2012; Accepted November 8, 2012; Published January 8, 2013 Copyright: ß 2013 Irandoust et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was supported by KiKa, Children Cancer Free(stichting Kinderen Kankervrij). website: http://kika.nl/. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected] . These authors contributed equally to this work. " These authors also contributed equally to this work.

Introduction diagnosed pediatric AML is achieved on more than 80% of patients, however, about 30–50% of these children relapse from Currently only one third of adult patients diagnosed with acute minimal residual disease (MRD) cells that apparently survived myeloid leukemia (AML) can be cured despite aggressive chemotherapy [4,5,6]. Therefore, new treatment modalities for chemotherapy, and relapse rate is still high in these patients AML are warranted. [1,2,3]. Although the prognosis of pediatric AML patients is Distinct morphological subgroups in French-American-British better, the outcome remains relatively poor. With standard (FAB) classification associate with different chromosomal rear- induction chemotherapy, complete remission (CR) for newly rangements and acquisition of recurring genetic abnormalities; for

PLOS ONE | www.plosone.org 1 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia example t(8;21))(q22;q22) and t(15;17))(q22;q21) create fusion were used in this study: ED9 (anti-rat SIRPa; mouse IgG1 isotype) genes, AML/ETO and PML/RARa, which predominate in FAB was labeled with Alexa-633, which was obtained from Invitrogen M2 and M3 AML subtypes respectively. These proteins are two of (Breda, The Netherlands). Considering the differences between rat the most common AML-associated oncofusion proteins, which in and human SIRPa, it is not likely that the ED9 agonistic antibody total represent 20% of the AML occurrence [7]. The cytogenetic cross-reacts with the human SIRPa [30,31]. For the experiments rearrangement involved with t(8;21) disrupt genes that are in the current study we used a concentration of 10 mg/ml ED9 required for normal hematopoietic development, such as subunits antibody. This concentration may be considered as relatively high, of core-binding factor [8]. Expression of the PML/RARa fusion but at this concentration the antibody is still specific since the protein leads to a differentiation block at the promyelocytic stage negative controls (i.e. empty vector cells) show no response to this that can be relieved by all-trans-retinoic acid (ATRA). Studies on antibody at all. A dose-response curve is depicted in Figure S1. AML/ETO and PML/RARa expressing cells have revealed that Rabbit polyclonal Ab8120 (Abcam, Cambridge, United King- aberrant signaling pathways are involved [9]. Insights into these dom) is directed to the cytoplasmic tail of human SIRPa. Mouse signaling pathways in AML at molecular level will pave ways for anti-actin monoclonal antibody, mAb1501R, was purchased from new treatment modalities. Chemicon International (Temecula, CA, USA). mAb against Signal regulatory protein alpha (SIRPa) is a transmembrane caspase-3 was obtained from Cell Signaling Technology (Boston, receptor composed of 3 immunoglobulin-like domains in its MA, USA). APC-labeled anti-human CD11b was acquired from extracellular region and an intracellular domain containing BD pharmingen (San Jose, CA, USA). PE- labeled B6H12 was immunoreceptor tyrosine-based inhibitory motifs (ITIMs) which purchased from Santa Cruz Biotechnology and B6H12 F(ab’)2- recruit and activate SHP-1 and SHP-2 [10,11]. SIRPa is fragments were generated in our laboratory by pepsin digestion as predominantly expressed on myeloid and neuronal cells [10] previously described [32]. and its activation has been implicated in regulation of different The histone deacetylase (HDAC) inhibitors (Trichostatin A cellular functions such as adhesion, migration, growth and (TSA), valproic acid (VPA) and sodium butyrate) and ATRA were differentiation [12,13,14]. Despite restricted expression of SIRPa, purchased from Sigma Aldrich (St Louis, MO, USA). 5-aza-2- CD47, the natural ligand for SIRPa [15], is ubiquitously expressed deoxycytidine (DAC, decitabine) was kindly provided by Pharma- and interacts with the SIRPa extracellular region. This interaction chemie BV (Haarlem, The Netherlands). Cytarabine (CytosarH) results in inhibition of phagocytosis by macrophages through was obtained from Pharmacia & Upjohn (Woerden, The Nether- tyrosine phosphatase activation and inhibition of myosin accumu- lands). Daunorubicin (CerubidineH) was purchased from Rhone lation [16,17,18]. CD47 functions as a ‘‘don’t eat me’’ signal and Poulenc Rorer (Amstelveen, The Netherlands). Etoposide (PV16, plays a key role in the programmed cell removal of abberant VepesidH) was obtained from Bristol-Myers Squib (Woerden, The versus normal cells [19]. Indeed, it was recently shown that CD47 Netherlands). Imatinib was provided by Novartis (The Nether- is overexpressed on AML leukemic stem cells as compared to their lands). zVAD was obtained from Merk Biosciences (Darmstadt, normal counterparts (hematopoietic stem cells) and this contrib- Germany). utes to inhibition of phagocytosis and clearance of LSCs [20]. In addition, blocking antibodies directed against CD47 promoted Patient samples phagocytosis as suggested through disruption of CD47-SIRPa For the expression array experiments bone marrow and/or interaction and this enhanced tumor clearance in vivo [20,21,22]. peripheral blood samples were collected from adult AML patients These findings are in line with several studies, which reported at diagnosis, as described by Valk PJ et al. [33]. Bone marrow elimination of tumor cells by employment of CD47 blocking and/or peripheral blood samples from children diagnosed with de antibodies [20,22,23,24,25,26,27]. novo AML were collected from the following study centers: VU Although SIRPa is known to be expressed by AML cells as well University Medical Center, Amsterdam, The Netherlands; The [11,28], its function on these cells has not been identified. Dutch Childhood Oncology Group (DCOG), The Hague, The Furthermore, previous studies were performed with antibodies Netherlands and the AML BFM-study Group, Hannover, that not only recognized SIRPa, but also several of the related Germany. AML subtypes were classified according to the criteria molecules such as SIRPb1 and SIRPc. In the present study we by Bennett et al., including the modifications to diagnose FAB have examined the mRNA and protein expression of SIRPa in a subtypes [34]. Mononuclear cells were isolated by density gradient large cohort of AML patients and determined its relevance for centrifugation as described previously [35]. All samples contained AML cell survival. We show for the first time that SIRPa ligation at least 80% leukemic cells, as determined morphologically by triggers programmed cell death in AML cells and synergizes with analyzing May-Gru¨nwald-Giemsa (Merck, Darmstadt, Germany) antileukemic agents. stained cytospins.

Design and Methods Cell Lines and culture conditions The human leukemic cell lines KG1a (primitive human Antibodies and drugs hematopoietic myeloid progenitor), Kasumi-1 (human acute At this moment no human agonistic SIRPa antibody is myeloid leukemia, FAB M2 t(8;21)), HL-60 (human promyelocytic available. However, a rat agonistic SIRPa antibody (ED9) was leukemia), NB4 (human acute promyeloctic leukemia, FAB M3 generated in our laboratory [10,29], that is also commercially t(15;17)),U937 (human acute monocytic leukemia), THP-1 available at Serotec (Oxford, UK). Such an agonistic antibody has (human acute monocytic leukemia), CEM (human acute lympho- much higher affinity to rat SIRPa as compared to CD47-Fc blastic leukemia), Jurkat (human T-cell acute lymphoblastic fragments [30,31], so the use of this agonistic antibody was leukemia) were routinely cultured in RPMI 1640 medium (Gibco preferred for mechanistic studies and optimal SIRPa triggering. Laboratories, Irvine, UK) supplemented with 10% fetal calf serum To be able to exert an agonistic signal using the available rat (Integro BV, Dieren, the Netherlands). Kasumi-1 cells (0.56106) antibody, a chimeric construct of SIRPa was generated carrying were incubated for 0, 3, 24, 48, 72 and 96 hours with 1 mM DAC, rat SIRPa extracellular domain and human transmembrane and 0.3 mM TSA, 0.5 mM VPA and 1 mM sodium butyrate in 5% cytoplasmic region. The following monoclonal antibodies (mAb) CO2 humidified air at 37uC. Cells were subsequently used for

PLOS ONE | www.plosone.org 2 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

Western Blot analysis, as described below. Human neutrophils EV were used as controls in all experiments. Ectopic expression of were isolated from heparinized blood of healthy individuals by SIRPa and/or eGFP were regularly monitored and were found to centrifugation over isotonic Percoll (Pharmacia, Uppsala, Sweden) be stable for several months, with .90% positive cells. and subsequent lysis of erythrocytes as described [36]. Neutrophils were cultured in Hepes-buffered saline solution supplemented with Cell proliferation and programmed cell death 1% human serum albumin (Cealb: Sanquin, Amsterdam, the For cell proliferation assays cells were seeded in triplicate at Netherlands) and 5 mM glucose. 16105/ml concentration in 96-well plates and treated with 10 mg/ ml of the ED9 mAb where indicated and incubated for the DNA isolation mentioned time points up to 7 days. For programmed cell death 6 Cell lines. 1–56106 of KG1a, Kasumi-1 and HL-60 cells (PCD) experiments, cells were seeded at 0.5610 cells/ml and on were resuspended in HIRT buffer (0.6% SDS, 10 mM Tris, days 1 and 3 PCD was measured by Annexin V-phycoerythrin 10 mM EDTA pH 8). Proteinase K was added and samples were and 7-amino-actinomycin D (7-AAD) or DAPI double staining incubated at 50uC for 2 hours and subsequently at 37uC according to the manufacturer’s protocol (BD Pharmingen, San overnight. Phenol: chloroform (1:1) was added and the solution Jose, CA, USA). All analyses were performed on a FACS calibur was mixed vigorously and centrifuged. The aqueous layer was (BD Biosciences, San Jose, CA, USA). Cells positively stained with transferred into a tube and the phenol/chloroform extraction was Annexin-V and 7-AAD negative were considered to be early repeated. Following centrifugation, the aqueous layer was apoptotic. removed and the DNA was precipitated by NaOAC/EtOH (1:24), washed with 70% EtOH and resuspended in TE buffer. Growth inhibition studies DNA concentrations were measured with spectrophotometer Growth inhibitory effects of chemotherapeutics in combination (Nanodrop, Isogen, The Netherlands). with ED9 mAb were evaluated with the MTT-assay, as described Patient samples. DNA was isolated from cryopreserved previously [39]. SIRPa and EV cells were incubated with 4 cytospins. A sterile swab and a drop of sterile water were used to concentrations of cytarabine (ara-C) 2.5–0.002 mM, daunorubicin wipe the cells from the slides. The swab was transferred into buffer (DNR) 18.0–0.005 mM, etoposide (VP16) 4.4–0.01 mM and (100 mM Tris, 10 mM NaCl, 5 mM EDTA, 1% SDS pH 9) imatinib 520.005 mM in combination with one fixed concentra- containing proteinase K and incubated overnight at 52uC. Tubes tion of the ED9 mAb (10 mg/ml); these experiments were done in were centrifuged and DNA was isolated as described above. For triplicate. Within each experiment all drugs were tested alone, as the method of the bisulphate sequencing of the DNA see Methods well as in combination. Drug interactions between chemothera- S1. peutic drugs and ED9 mAb were studied by using the multiple drug effect analysis of Chou and Talalay [40] (Calcusyn software, Western blot analysis Biosoft, Cambridge, UK) and antagonistic, additive or synergistic Cells were washed in PBS, centrifuged and the cell pellet was interactions were determined. This method is commonly used in lysed with Igepal lysis buffer (Sigma-Aldrich) containing protease many drug interaction studies [41]. The interactions are deter- inhibitor cocktail (Roche). Whole cell lysates were clarified by mined by Combination Index (CI) which indicates synergism , . centrifugation and denatured in Laemmli’s sample buffer (Bio-rad (CI 0.9), additivity (CI = 0.9–1.1) or antagonism (CI 1.1). In the . Laboratories, Hercules, CA, USA). After that cell lysates were CI-FA plot the CI values 0.5 are evaluated and per experiment a subjected to Western blot and membranes stained with primary mean CI was calculated from FA values 0.5, 0.75 and 0.9. The average CI (6 SD) of three experiments is given for each of the and secondary antibodies. combinations. Construction of retroviral vectors and transduction of Statistical analysis Kasumi-1 cells The statistical significance of measured differences in prolifer- The rat-human SIRPa fusion construct (chSIRPa) was gener- ation and PCD between the various conditions and cell ated from cDNA and PCR fragments as follows: nucleotide 1– populations was determined using the paired Student’s t-test. 1236 of the rat SIRPa cDNA [10] was fused to nucleotide 1230– Calculations were performed using Graph-pad Prism and SPSS 1509 of the human cDNA (prot. accession No: NM_080792). The software. Statistical analysis (Cox proportional hazards model; chSIRPa protein contains amino acids 1–412 (rat extracellular reported p values corresponded to the Wald test) on overall domain) and amino acids 411–503 (human transmembrane and survival and event free survival was performed in SPSS software. cytoplasmic region) resulting in a total length of 505 amino acids, Survival distribution was compared with median SIRPa expres- including the signal sequence. The sequence of the construct was sion of the whole group. confirmed by automated sequencing. For retroviral transduction the chSIRPa construct was cloned Results into the retroviral expression vector pLZRSpMBN-linker-IRES- eGFP(NotI) [37]. The Phoenix-A packaging cell line [38] was SIRPa mRNA expression in AML transfected with the retroviral construct containing chSIRPa or The mRNA level in pediatric AML patients was analyzed in a empty vector (EV) containing eGFP by calcium-phosphate micro-array dataset containing 226 samples [42]. SIRPa mRNA transfection. After puromycin selection (Sigma-Aldrich, St Louis, expression varied considerably among different AML patients MO, USA), harvested virus supernatant [37] was used for (Figure 1A). A clear association was observed between SIRPa transduction of Kasumi-1 or NB4 cells. chSIRPa-expressing cells expression and AML FAB subtypes with the highest levels found in were subsequently selected in several rounds by FACS sorting the myelo-monocytic FAB M4/M5 subsets. The SIRPa expression (MoFlo,Dako Cytomation) on the basis of eGFP expression to levels in myeloblastic leukemic blasts were relatively low in the reach to .98% positive cells. Cell surface expression of chSIRPa FAB M0–M3 subtypes. The M6 erythroid type of AML also was determined by FACS analysis using the ED9 mAb, as showed a low SIRPa expression. Comparing acute myeloid described below. FACS-sorted, mock-transduced cells containing leukemia cases with normal bone marrow, lower expression of

PLOS ONE | www.plosone.org 3 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

SIRPa was observed in immature AML subtypes (Figure 1A), protein expression was low/undetectable in immature subgroups suggesting a myeloid differentiation stage-dependent expression, compared to the more mature groups such as M4 and M5 which is in line with the high expression of SIRPa found on (Figure 2B, C). As expected we did not observe SIRPa expression normal monocytes and macrophages [10]. In particular, the in ALL patient samples (n = 10) (Figure S3). Collectively, these significant difference between M0–M3 and M4/M5 (p,0.001), findings suggest a selective myeloid and differentiation stage- and the different expression distribution are depicted in Figure 1B. dependent expression of SIRPa mRNA and protein expression in Classifying pediatric samples based on karyotypes showed the AML. highest (p = 1.63610210) and lowest (p = 2.0561026) expression of SIRPa in MLL-rearranged and t(8;21) subgroups, respectively Upregulation of SIRPa upon differentiation of t(15;17) (Figure 1C). Since we only accessed one pediatric dataset, we AML cells validated these findings in adult AML datasets [33,43,44] and To address whether SIRPa expression is upregulated upon consistent with the pediatric results the highest level of SIRPa differentiation of AML cells, we selected the NB4 cell line, a expression was found in M4 and M5 subtypes as compared to the t(15;17) M3 FAB subtype, which only express low levels of SIRPa immature groups such as FAB M0, M1, M2 and M3 (Figure S2). (Figure 2A). Since ATRA treatment of t(15;17) APL patients is In addition, karyotype classification of 285 AML patients in an known to result in granulocytic differentiation [47,48], we adult dataset [33], showed increased SIRPa expression in inv(16), examined if SIRPa expression increased after exposure to ATRA. and MLL-rearrangement groups (depicted as clusters 5, 9 and 16) in comparison to t(8;21) and t(15;17) AML (Figure S1B). To address this, the NB4 cells were incubated with 1 mM ATRA for 7 consecutive days and granulocytic differentiation of the NB4 To examine whether SIRPa expression is correlated with cells was confirmed by upregulation of the common myeloid patient survival, we performed an analysis on overall survival (OS) and event free survival (EFS) in the pediatric cohort (n = 175), for marker CD11b (Figure 3A). In concert with the increased a which follow-up data are available. We observed that higher differentiation of NB4 cells, SIRP protein expression was SIRPa expression compared to the median of the 175 patients (8.1 markedly upregulated following ATRA exposure (Figure 3B). arbitrary units), significantly correlated with unfavorable outcome. This upregulation was already detectable after 24 hrs and it was Figure 1D shows the Kaplan Meier analysis based on OS (log-rank further increased during the following days of treatment. p = 0.024, hazard ratio (HR) 1.7, p = 0.026). For EFS the data were similar: log-rank p = 0.029, and HR: 1.5, p = 0.031. In Induction of programmed cell death in t(15;17) AML cells addition, following stratification on karyotype, we could not following SIRPa ligation generally find any significant relation between SIRPa and Our initial experiments with the rat myeloid cell line NR8383 outcome (not shown). Only within the MLL rearranged samples showed that several agonistic monoclonal antibodies [49] against a high SIRPa expression (above the median SIRPa expression of rat SIRPa (for example ED9, ED17 or OX41) or recombinant Fc- the MLL rearranged group) exhibited a trend towards unfavorable fusion proteins containing the extracellular region of CD47, the outcome (HR: 2.28, p = 0.062). natural ligand of SIRPa, are able to suppress cell proliferation It is difficult to compare the outcome between children and (Figure S4). In order to investigate the effect of SIRPa ligation in adults since the adults have, in general, a very dismal prognosis. In NB4 cells we tested a variety of previously reported antibodies the data set of Valk et al. [30] no association was found between against SIRPa, but these either lacked the appropriate specificity, SIRPa expression and outcome on the whole cohort. For the adult showing cross-reactivity with other SIRP family members, or MLL rearranged samples, high SIRPa levels associated with a lacked the agonistic activity [28,32]. In addition CD47-Fc did not slightly favorable outcome (HR: 0.84, p = 0.027). In both children show a sufficiently high affinity for binding to SIRPa in our in vitro and adults, multivariate analysis reveals that SIRPa is not an experiments to be used as an agonistic (data not shown). Hence, to independent risk factor. be able to study the effect of SIRPa triggering in human myeloid Recent studies have shown that CD47, the ligand for SIRPa,is cells, we generated a chimeric SIRPa (chSIRPa) construct that a prognostic factor in breast cancer and its expression correlates enabled the use of the rat specific SIRPa agonistic ED9 mAb. This with SIRPa expression in bone marrow and peripheral blood of chSIRPa construct consisted of the extracellular region of rat breast cancer patients [32,45]. We therefore evaluated a possible SIRPa and the transmembrane and the cytoplasmic domains of association between SIRPa and CD47 expression in different human SIRPa [10]. Stable t(15;17) NB4 cell lines expressing datasets [33,42,46], however we did not find any evidence for such chSIRPa or empty vector (EV) were generated by retroviral an association in AML. Figure 1E shows that CD47 is equally transduction. Flow cytrometric analysis of the retrovirally trans- distributed among the pediatric AML subtypes while there is a duced and FACS-sorted cells showed that the vast majority clear difference in SIRPa expression with the MLL-rearranged (.90%) of cells had been transfected by chSIRPa (Figure 3C). clustering in the high SIRPa range. The levels of chSIRPa expression (i.e. mean fluorescence) were comparable to those seen, with the same mAb, on rat SIRPa protein expression in AML macrophages or granulocytes ([10] and data not shown). We determined SIRPa protein expression, by Western blotting Ligation of SIRPa in NB4 cells by agonistic ED9 mAb resulted using an antibody directed to the cytoplasmic tail of the human in induction of programmed cell death (PCD) as quantified by flow SIRPa on various leukemic cell lines and patient samples. While cytometry using annexin-V/7-AAD staining (Figure 3C). After no expression was observed among acute lymphoblastic leukemia 24 h of exposure to ED9 mAb, the percentage of annexin-V (ALL) cell lines, AML cell lines differentially expressed the SIRPa positive cells was significantly higher in the NB4 chSIRPa cells protein (Figure 2A). In particular, immature myeloblasts such as (47.368.6%), as compared to NB4 EV cells (15.165.0%; t(8;21) Kasumi-1, KG-1, HL60 cells or promyelocytes like t(15;17) p = 0.009) (Figure 3C). These data support the requirement for NB4 cells expressed low or undetectable levels of SIRPa protein ED9 binding to SIRPa, since no induction of cell death was compared to more differentiated monocytic cells such as THP-1 observed in NB4 EV cells. These findings provide evidence for and U937 (Figure 2A). We also analyzed 20 primary pediatric induction of cell death capacity by SIRPa triggering in APL cells. AML patient samples and consistent with the mRNA data, SIRPa

PLOS ONE | www.plosone.org 4 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

Figure 1. SIRPa mRNA expression and its prognostic effect in pediatric AML cohort. (A) SIRPa mRNA expression was determined in different FAB subtypes. The dots represent individual patients and the horizontal bar is the group mean. The horizontal dotted line represents the mean levels (139.6; n = 5) of SIRPa expression in normal CD34+ HSC. (ND: not determined). (B) Frequency of SIRPa expression among different FAB subtypes of AML patients is shown as stacked histograms. (C) SIRPa mRNA expression as stratified after karyotype (NK: normal karyotype). (D) Overall survival of pediatric AML patients (n = 175) stratified according to either low (, median of 8.1) or high ($ median 8.1) SIRPa mRNA expression. (E) Correlation between CD47 and SIRPa mRNA expression is shown by the blue and red dots, respectively. The lower bar demonstrates the cluster of patients in karyotypes, in which higher SIRPa expression is clustered in blue, representing the MLL rearrangement group on the right side. doi:10.1371/journal.pone.0052143.g001

PLOS ONE | www.plosone.org 5 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

PLOS ONE | www.plosone.org 6 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

Figure 2. SIRPa protein expression in AML cell lines and patients. Western blot analysis was performed in (A) cell lines and (B) 20 pediatric AML patient samples. b-actin staining was used as loading control. (C) SIRPa expression is quantified relative to b-actin expression. doi:10.1371/journal.pone.0052143.g002

From a therapeutic perspective, it would appear beneficial to Inhibition of proliferation and induction of PCD in t(8;21) relieve differentiation block in M3 AML cells by using ATRA and AML cells by SIRPa triggering as a result to upregulate SIRPa expression, which can subse- Since triggering SIRPa by agonistic ED9 mAb induced cell quently be targeted by an agonistic antibody to induce PCD. death in t(15;17) NB4 cells, we extended our findings in t(8;21) Clearly prerequisite for such a strategy to be successful would be Kasumi-1 cells, stably transduced with chSIRPa or EV (Figure 5A). the efficiency of SIRPa triggering following differentiation. The overexpression of chSIRPa in Kasumi-1 cells itself did not Therefore we examined whether apoptosis induction via SIRPa affect the growth, however culturing Kasumi-1 chSIRPa cells in persisted after differentiation with ATRA. Stably transduced NB4 the presence of the agonistic rat ED9 antibody caused a significant (chSIRPa and EV) cells were exposed to 1 mM ATRA in inhibition of proliferation (Figure S7). In order to investigate combination with the fixed concentration of 10 mg/ml ED9 whether growth suppression triggered by chSIRPa ligation mAb, which was shown to trigger PCD in the NB4 chSIRPa cells. coincided with an enhanced level of cell death induction, the As expected, ATRA treatment alone did not have any effect on percentage of cell death was quantified by flow cytometry, using PCD, whereas ED9 resulted in PCD in the chSIRPa transduced Annexin V and 7-AAD staining (Figure 5B). Already 24 hours NB4 cells and this was not significantly altered after differentiation after adding ED9 mAb the percentage of early dying cells, defined with ATRA. Furthermore while SIRPa triggering by ED9 mAb as Annexin V positive, was significantly higher in the Kasumi-1 does induce programmed cell death, we found that it did not affect chSIRPa cells, as compared to Kasumi-1 EV cells, while addition differentiation (Figure S5). of an irrelevant antibody had no effect (Figure 5B, lower panel). Taken together, these data indicate that ATRA provides a Similar results were observed on day 3 of treatment, at which stimulus for differentiation of t(15;17) APL cells and this results in ligation with ED9 mAb had caused significant cell death in upregulation of SIRPa expression to a level that makes the cells Kasumi-1 chSIRPa cells (35.2615.3% versus 10.462.8% in prone to cell death induction via SIRPa triggering even in untreated control cells; p = 0.02). All effects required ED9 binding differentiated M3 cells. to chSIRPa, since no induction of cell death was observed in Kasumi-1 EV cells. Upregulation of SIRPa following differentiation of t(8;21) We next investigated whether cell death induction in Kasumi-1 AML cells cells was involved caspase activity, which is often required for PCD. First, we investigated activation of the effector caspase-3, To examine whether the low endogenous expression of SIRPa which can be measured by evaluating the appearance of the p17 is also upregulated following differentiation of other low-SIRPa- caspase-3 cleavage product. As shown in Figure 5C triggering of expressing myeloid leukemic cells, we selected t(8;21) Kasumi-1 SIRPa in Kasumi-1 chSIRPa cells did not result in any detectable cells. These cells belong to AML M2 FAB subtype and express low caspase-3 cleavage, whereas this was detected upon culture of endogenous levels of SIRPa (Figure 2A). It has been shown that freshly isolated neutrophils. Furthermore, incubation with the histone deacetylase (H-DAC)-inhibitors such as butyrate, valproic universal inhibitor of caspases, zVAD (10 mM), did not affect acid (VPA) and trichostatin (TSA) and DNA methyltransferase programmed cell death induction via SIRPa in Kasumi-1 (DNMT)- inhibitors such as decitabine, induce granulocytic chSIRPa cells whereas neutrophil apoptosis was zVAD sensitive maturation of t(8;21) acute myeloid leukemia cells [50,51,52,53] (results not shown). Collectively, these findings indicate a growth- (and data not shown). suppressive and caspase-independent mode of PCD induction via Kasumi-1 cells were exposed to TSA, VPA,butyrate or SIRPa in t(8;21) AML. decitabine for indicated time points, which resulted in markedly To examine whether the observed effects of ED9 mAb in increased SIRPa levels (Figure 4). With all drugs, an increase in Kasumi-1 cells, occur through blocking of CD47-SIRPa interac- SIRPa protein expression was detected as early as 3 hrs after tions, we used the blocking anti-CD47 antibody B6H12. As shown exposure and this expression reached maximal levels after in Figure S8, induction of cell death by ED9 cannot be mimicked approximately 24 hrs. These data show that SIRPa is upregulated by B6H12. This experiment shows at least that the pro-apoptotic/ following differentiation of Kasumi-1 cells. growth regulatory effects that we report in this study are not simply An alternative explanation for the upregulation of SIRPa in due to a blocking of cis or trans CD47-SIRPa interactions and are t(8;21) AML was that these inhibitors of epigenetic silencing had more likely due to agonism of the ED9 antibody that was actually acted directly on the SIRPa gene (accession number: NP- also reported by us before in another context [54]. 542970.1), and in fact this seemed possible since a prominent CpG island is present in the PTPNS1 promoter region. DNA SIRPa ligation synergizes with conventional antileukemic methylation in this region was explored in Kasumi-1 cells and four and targeted agents in t(8;21) and t(15;17) AML cells t(8;21) AML patients by bisulphite DNA sequencing. Results Considering the potential of exploiting SIRPa targeting to revealed actually very low levels of DNA methylation in the improve the treatment of AML patients, we examined the efficacy promoter region (Figure S6). We also analyzed the SIRPA2p of the ED9 mAb in combination with clinically relevant pseudogene, in which abundant methylation was detected in the chemotherapeutic agents used for the treatment of AML. NB4 corresponding region [49]. Taken together these data strongly chSIRPa and EV cells were exposed to cytarabine (ARA-C) and suggest that the increased level of SIRPa in t(8;21) Kasumi-1 cells daunorubicin (DNR) in combination with ED9 mAb. Survival was following demethylating agents is the result of differentiation. monitored after 4 days using a range of chemotherapeutic drug concentrations, which had shown appropriate dose response curves in pilot experiments. We used a fixed concentration of

PLOS ONE | www.plosone.org 7 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

and PE-7AAD FACS staining. (E) Percentage of apoptosis after exposure to 1 mM ATRA is shown in combination with 10 mg/ml of ED9. doi:10.1371/journal.pone.0052143.g003

ED9 mAb (10 mg/ml), which had been shown to promote PCD in NB4 cells (Figure 6A). Co-incubation of NB4 cells stably expressing chSIRPa with each of the two chemotherapeutics and ED9 mAb resulted in synergistic effects (combination indexes (CI) using standard calcusyn calculation were CIDNR = 0.0.5960.04 and CIARA-C = 0.6060.2). No effect of ED9 mAb was seen in the NB4 EV cells (data not shown). In addition, Kasumi-1 chSIRPa and EV cells were exposed to ARA-C, DNR, and VP16 in combination with ED9. Since the Kasumi-1 cell line has been described to harbor an activating c-kit mutation [55], we also tested ED9 in combination with the tyrosine kinase inhibitor imatinib mesylate (Figure 6B). Similar to the experiments with the NB4 cell line, this was tested using a 4- day survival assay applying a range of drug concentrations, which had shown appropriate dose response curves in pilot experiments, and a fixed concentration of ED9 mAb (10 mg/ml). The leukemic cell survival of Kasumi-1 EV and chSIRPa cells after incubation with ED9 alone were 9862.4% and 8764.4%, respectively. Co- incubation of Kasumi-1 SIRPa cells with each of the four antileukemic drugs and ED9 resulted in synergism which was indicated by a shift in the dose-response curve. By using the standard Calcusyn calculations for combination effects, a synergistic effect of ED9 incubation was observed for all applied chemotherapeutic drugs (combination indexes (CI) include: CIARA-C = 0.4660.32; CIDNR = 0.7460.06, CIVP16 = 0.5560.066 and CIImatinib = 0.7560.11) (Figure 6B). Taken together, these results show that expression and ligation of chSIRPa provide growth inhibitory effects in t(15;17) and t(8;21) AML cells and this had a synergism with established antileukemic drugs.

Discussion Insights into the molecular pathogenesis of AML have paved the way for new treatment strategies that specifically target gene products implicated in induction of the leukemia. In the present study we have investigated the role of SIRPa as a potential target in the treatment of AML. By evaluating SIRPa mRNA levels in both pediatric and adult cohorts of AML patients, we observed a differential expression of SIRPa in AML subtypes. Interestingly, high expression of SIRPa was observed in more mature AML subgroups (M4 and M5) in comparison to immature subtypes, normal bone marrow and CD34+ blast cells. There was interpatient vairability of SIRPa expression between patients with M4 or M5 subtypes, which is probably due to heterogeneity of AML. Nevertheless, higher expression in mature subtypes is consistent with a differentiation related expression of SIRPa. Consistent with this, an upregulation of SIRPa was also observed during differentiation of AML cell lines, which express low endogenous levels of SIRPa. Figure 3. Upregulation of SIRPa upon differentiation of The expression of SIRPa was not correlated to the expression of t(15;17) NB4 cells and induction of cell death following its triggering. (A) NB4 cells were exposed to 1 mM ATRA and granulocytic its ligand CD47, which was ubiquitously expressed on the AML differentiation of the cells was examined by cell surface expression of blasts. This is consistent with a recent study performed by the common myeloid marker, CD11b. (B) SIRPa protein expression, Nagahara et al. who showed that increased expression of SIRPa determined by western blotting, is upregulated in ATRA-incubated NB4 and CD47 was not correlated on breast cancer cell lines. However cells. b-actin is used as a loading control. (C) Flow cytometric analysis of a stronger correlation was observed in the bone marrow and chSIRPa surface expression is determined by using ED9 mAb in transduced NB4 empty vector and chSIRPa expressing cells. (D) 24 hrs peripheral blood of breast cancer patients compared to normal following ED9 (10 mg/ml) incubation, the percentage of cell death in cases [45]. They suggest that such host factor characteristics may chSIRPa and EV transduced NB4 cells was quantified by APC-Annexin V have implications for prognosis of breast cancer. We also evaluated mRNA levels of other related genes in the various datasets of

PLOS ONE | www.plosone.org 8 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

Figure 4. SIRPa upregulation in t(8;21) Kasumi-1 cells following treatment with inhibitors of epigenetic gene silencing. Kasumi-1 cells were incubated with 1 mM Decitabine, 0.5 mM valproic acid, 1 mM Butyrate and 300 nM Trichostatin. Endogenous SIRPa protein level, determined by Western blotting was upregulated at indicated time points. b-actin staining was used as a loading control. doi:10.1371/journal.pone.0052143.g004

AML. No clear association was observed between CD47 and the hematopoietic phosphatase SHP-1 can indeed be recruited SIRPa or different FAB subtypes or karyotypes. In our AML and activated upon triggering of SIRPa by its natural ligand study, the association between SIRPa and prognosis is not strong CD47, as well as by the agonistic antibody ED9 used herein [54]. and does not function as an independent factor. Galbaugh et al It should be emphasized that while the survival and proliferation showed that SIRPa mRNA was high in triple negative breast of AML may not be directly regulated by CD47-SIRPa cancer and related to an increased invasiveness of the tumor [56]. interactions, there the in vivo life span of leukemic cells may well However, thus far there are no publications that show a clear be affected by them in another way. In particular, recent studies correlation between SIRPa expression and outcome, even for have demonstrated that CD47 can act as an anti-phagocytic or so breast cancer. The lack of differential expression of CD47 in our called ‘‘don’t eat me’’ signal that prevents clearance of human study might be due to the fact that we investigated the bulk of the leukemic cells by macrophages in xenogeneic mouse models in vivo AML cells and not the leukemic stem cells, for which it was [20,21]. Other ‘‘don’t eat me’’ signals such as CD200 have also previously established that CD47expression is associated with poor been shown to be upregulated in multiple tumors including AML prognosis [20]. [58]. Leukemic stem cells appear to have higher levels of CD47 We demonstrate here for the first time, that SIRPa ligation than normal CD34+ HSC cells and this could provide them with a using agonistic monoclonal antibodies inhibits cell growth and selective advantage for survival. In this study we show that other promotes cell death induction. Our results suggest that PCD via signals such as SIRPa can be upregulated to encourage such don’t SIRPa is caspase-independent. This mechanism of cell death eat me signals and subsequent evasion from programmed cell induction has been reported previously in the context of AML removal [19,58]. It must be noted that CD47 signaling can also be [57], in which no involvement of caspases in cell death of an AML regulated through binding to its ligand, thrombospondin-1(TSP- cell line was observed upon treatment with chemotherapeutic 1). This CD47-TSP-1 interaction has been shown to inhibit drugs. It should be noted that CD47 ligation also induces caspase- response to nitric oxide and correspondingly increase radiosensi- independent PCD [57], and it would seem reasonable to assume tivity. As a result, blocking such interactions could confer that this was due to a potential blocking of cis-interactions between therapeutic radioprotection of normal tissues [59,60,61]. CD47 and SIRPa on AML, rather than to agonistic triggering of Analysis of AML patients had revealed that higher levels of both receptors. However, observations show that this was not CD47 are associated with a poor prognosis [17,18]. While the likely to be the case, since the effects on t(15;17) NB4 cells and results of these studies indicate that blocking of the interaction t(8;21) Kasumi-1 cells could not be reproduced using blocking between CD47 and SIRPa may be of interest from a therapeutic antibodies against CD47 (Alvarez et al, not shown). The actual perspective, our current results suggest that it may, perhaps even mechanism underlying the pro-apoptotic effect of SIRPa trigger- simultaneously be beneficial to trigger SIRPa as well. Especially ing is under investigation. Clearly, one obvious candidate to since targeting the CD47–SIRPa phagocytic pathway alone is mediate growth inhibitory signals, particularly in myeloid cells, is likely to have toxic effects [58]. In fact, the ED9 antibody against the cytosolic tyrosine phosphatase SHP-1. Our preliminary SIRPa that we have used herein appears both capable of findings show that SHP-1 and also the related SHP-2 are in fact triggering programmed cell death as well as to block CD47- abundantly expressed in fresh AML samples and in established SIRPa interaction [62]. Clearly, future studies are needed to AML cell lines (data not shown). In rodent macrophage cell lines

PLOS ONE | www.plosone.org 9 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

PLOS ONE | www.plosone.org 10 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

Figure 5. Ligation of chSIRPa induces caspase 3-independent PCD in Kasumi-1 cells. (A) Flow cytometric analysis of SIRPa expression was performed by using ED9 mAb in stable Kasumi-1 cells expressing chSIRPa and EV. (B) Kasumi-1 chSIRPa and EV cells were incubated with 10 mg/ml ED9 mAb and the percentage of cell death was determined after 24 hrs. Annexin V and 7-AAD FACS staining defined that ligation of chSIRPa resulted in increased cell death in chSIRPa Kasumi-1 cells compared to EV control cells. Data are means 6 SD calculated from 3 independent experiments using triplicate samples (*: significant difference p,0.05). (C) Kasumi-1 cells expressing chSIRPa or EV were treated with 10 mg/ml ED9 for mentioned time points. Caspase 3 staining shows no cleavage of the p32 subunit. As a positive control for caspase 3 cleavage, human neutrophils (PMN) were incubated at room temperature for 0 and 24 hours. doi:10.1371/journal.pone.0052143.g005

generate the suitable agents that can trigger apoptosis via human SIRPa and validate it as a potential treatment target in AML.

Supporting Information Figure S1 Dose-response curve of ED9 antibody induced apoptosis in Kasumi-1 cells. EV = kasumi cells tranduced with empty vector, WT = kasumi cells tranduced with wild type SIRPa. Apoptosis was measured after exposure to a range of ED9 antibody concentrations. 10 mg/ml was selected as optimal concentration for further studies. At this concentration no effect was seen on cells only expressing the (human) constitutive SIRPa. (PPT) Figure S2 SIRPa mRNA expression in adult AML cohort. (A) SIRPa mRNA expression was determined in different FAB subtypes of 285 adult patients. The dots represent individual patients and the horizontal bar is the mean of the group (ND: not determined). (B) Adapted correlation view of the 16 unsupervised clusters (indicated on the left) of 285 adult AML specimens identified by mRNA profiling [33], including the expression levels of SIRPa using 3 independent probes on the right diagonal axes. SIRPa expression is high in clusters 5, 9 and 16, but low in most other clusters, including clusters 12 and 13, which contain almost exclusively t(15;17) and t(8;21) AML, respectively. (PPT) Figure S3 SIRPa is not expressed in ALL patient samples. Analysis of protein expression of SIRPa in pediatric ALL patient samples by western blotting showed that SIRPa is not expressed in these samples. b-actin staining was used as a loading control. (PPT) Figure S4 Triggering SIRPa in the rat NR8383 macrophage cell line inhibits proliferation. NR8383 cells were incubated for 18 hours with CD47-Fc protein or indicated anti-rat SIRPa monoclonal antibodies (ED9, ED17 or OX41). 3H- thymidine was added for 4 hours and proliferation was determined by incorporated radioactivity. (PPT) Figure S5 NB4 cells differentiate by ATRA exposure. Differentiation of NB4 cells stably expressing chSIRPa and EV was examined by flow cytometry after treatment with ATRA or ED9. increased expression of CD11b was observed only after ATRA but not by ED9 treatment. (PPT) Figure S6 PTPNS1 promoter region is not methylated. Figure 6. SIRPa-derived signal synergizes with different Each circle indicates a CpG dinucleotide (open circles: unmethy- antileukemic drugs. Inhibition of cell growth is depicted by lated, filled circles: methylated) and each line represents analyses of combination of ED9 mAb (10 mg/ml) with (A) Ara-C and DNR in NB4 cells expressing chSIRPa (B) Ara-C, DNR, VP16, DAC and imatinib in a single amplified clone. SIRPa2p pseudogene, which is highly Kasumi-1 cells expressing chSIRPa. Results are based on means of 3 highly homologous to PTPNS1 was used as a positive control with experiments and are calculated using Calcusyn. high degree of methylation [49]. Methylation specific PCR and doi:10.1371/journal.pone.0052143.g006 bisulphate sequencing [63] of the Kasumi-1 cell line and four

PLOS ONE | www.plosone.org 11 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia t(8;21) AML patients did not reveal methylation of the PTPNS1 blocking anti-CD47 antibody. Percentage of cell death was promoter region. increased significantly in the case of ED9 treatment compared to (PPT) EV but B6H12 anti-CD47 incubation did not have this effect. Figure S7 SIRPa ligation results in inhibition of prolif- (PPT) eration in Kasumi-1 cells. Kasumi-1 cells expressing chSIRPa Methods S1 Detailed method description of the DNA bisul- or EV, were incubated with ED9 mAb for 7 days and cell phate sequencing. proliferation was evaluated by daily cell counting. Data are means (DOC) 6 SD calculated from 3 independent experiments using triplicate samples. Author Contributions (PPT) Provided patient samples and patients’ characteristic: ES PK UC DR EdB Figure S8 Blocking anti-CD47 antibody cannot mimic CMZ MvdHE. Statistical analysis: JC EC PV RD. Experimental ED9 effects in Kasumi-1 cells. (A) Flow cytometry data of performance: AB MA KS MI EvB IH JAZ. Conceived and designed the DAPI and Annexin-V staining and (B) Summary graph illustrates experiments: TvdB JC MI IH EvB GK AvdL. Performed the experiments: the quantified flow cytometric data. Kasumi-1 cells expressing AB MA KS MI EvB IH JAZ. Wrote the paper: MI JC TvdB IH JAZ. chSIRPa or EV were incubated with ED9 mAb or B6H12 as

References 1. Hann IM, Stevens RF, Goldstone AH, Rees JK, Wheatley K, et al. (1997) 21. Jaiswal S, Jamieson CH, Pang WW, Park CY, Chao MP, et al. (2009) CD47 is Randomized comparison of DAT versus ADE as induction chemotherapy in upregulated on circulating hematopoietic stem cells and leukemia cells to avoid children and younger adults with acute myeloid leukemia. Results of the Medical phagocytosis. Cell 138: 271–285. Research Council’s 10th AML trial (MRC AML10). Adult and Childhood 22. Chao MP, Alizadeh AA, Tang C, Jan M, Weissman-Tsukamoto R, et al. (2011) Leukaemia Working Parties of the Medical Research Council. Blood 89: 2311– Therapeutic antibody targeting of CD47 eliminates human acute lymphoblastic 2318. leukemia. Cancer Res 71: 1374–1384. 2. Goldstone AH, Burnett AK, Wheatley K, Smith AG, Hutchinson RM, et al. 23. Uno S, Kinoshita Y, Azuma Y, Tsunenari T, Yoshimura Y, et al. (2007) (2001) Attempts to improve treatment outcomes in acute myeloid leukemia Antitumor activity of a monoclonal antibody against CD47 in xenograft models (AML) in older patients: the results of the United Kingdom Medical Research of human leukemia. Oncol Rep 17: 1189–1194. Council AML11 trial. Blood 98: 1302–1311. 24. Chao MP, Alizadeh AA, Tang C, Myklebust JH, Varghese B, et al. (2010) Anti- 3. Rowe JM, Neuberg D, Friedenberg W, Bennett JM, Paietta E, et al. (2004) A CD47 antibody synergizes with rituximab to promote phagocytosis and phase 3 study of three induction regimens and of priming with GM-CSF in older eradicate non-Hodgkin lymphoma. Cell 142: 699–713. adults with acute myeloid leukemia: a trial by the Eastern Cooperative Oncology 25. Chan KS, Espinosa I, Chao M, Wong D, Ailles L, et al. (2009) Identification, Group. Blood 103: 479–485. molecular characterization, clinical prognosis, and therapeutic targeting of 4. Kaspers GJ, Zwaan CM (2007) Pediatric acute myeloid leukemia: towards high- human bladder tumor-initiating cells. Proc Natl Acad Sci U S A 106: 14016– quality cure of all patients. Haematologica 92: 1519–1532. 14021. 5. Ravindranath Y (2003) Recent advances in pediatric acute lymphoblastic and 26. Edris B, Weiskopf K, Volkmer AK, Volkmer JP, Willingham SB, et al. (2012) myeloid leukemia. Curr Opin Oncol 15: 23–35. Antibody therapy targeting the CD47 protein is effective in a model of aggressive 6. Hann IM, Webb DK, Gibson BE, Harrison CJ (2004) MRC trials in childhood metastatic leiomyosarcoma. Proc Natl Acad Sci U S A. acute myeloid leukaemia. Ann Hematol 83 Suppl 1: S108–112. 27. Willingham SB, Volkmer JP, Gentles AJ, Sahoo D, Dalerba P, et al. (2012) The 7. Martens JH, Stunnenberg HG (2010) The molecular signature of oncofusion CD47-signal regulatory protein alpha (SIRPa) interaction is a therapeutic target proteins in acute myeloid leukemia. FEBS Lett 584: 2662–2669. for human solid tumors. Proc Natl Acad Sci U S A. 8. Goyama S, Mulloy JC (2011) Molecular pathogenesis of core binding factor 28. Seiffert M, Brossart P, Cant C, Cella M, Colonna M, et al. (2001) Signal- leukemia: current knowledge and future prospects. Int J Hematol 94: 126–133. regulatory protein alpha (SIRPalpha) but not SIRPbeta is involved in T-cell 9. Alcalay M, Meani N, Gelmetti V, Fantozzi A, Fagioli M, et al. (2003) Acute activation, binds to CD47 with high affinity, and is expressed on immature myeloid leukemia fusion proteins deregulate genes involved in stem cell CD34(+)CD38(2) hematopoietic cells. Blood 97: 2741–2749. maintenance and DNA repair. J Clin Invest 112: 1751–1761. 29. Damoiseaux JG, Dopp EA, Neefjes JJ, Beelen RH, Dijkstra CD (1989) 10. Adams S, van der Laan LJ, Vernon-Wilson E, Renardel de Lavalette C, Dopp Heterogeneity of macrophages in the rat evidenced by variability in EA, et al. (1998) Signal-regulatory protein is selectively expressed by myeloid determinants: two new anti-rat macrophage antibodies against a heterodimer and neuronal cells. J Immunol 161: 1853–1859. of 160 and 95 kd (CD11/CD18). J Leukoc Biol 46: 556–564. 11. Seiffert M, Cant C, Chen Z, Rappold I, Brugger W, et al. (1999) Human signal- 30. Hatherley D, Harlos K, Dunlop DC, Stuart DI, Barclay AN (2007) The regulatory protein is expressed on normal, but not on subsets of leukemic structure of the macrophage signal regulatory protein alpha (SIRPalpha) myeloid cells and mediates cellular adhesion involving its counterreceptor CD47. inhibitory receptor reveals a binding face reminiscent of that used by T cell Blood 94: 3633–3643. receptors. J Biol Chem 282: 14567–14575. 31. Vernon-Wilson EF, Kee WJ, Willis AC, Barclay AN, Simmons DL, et al. (2000) 12. Fukunaga A, Nagai H, Noguchi T, Okazawa H, Matozaki T, et al. (2004) Src CD47 is a ligand for rat macrophage membrane signal regulatory protein SIRP homology 2 domain-containing protein tyrosine phosphatase substrate 1 (OX41) and human SIRPalpha 1. Eur J Immunol 30: 2130–2137. regulates the migration of Langerhans cells from the epidermis to draining 32. Zhao XW, van Beek EM, Schornagel K, Van der Maaden H, Van Houdt M, et lymph nodes. J Immunol 172: 4091–4099. al. (2011) CD47-signal regulatory protein-alpha (SIRPalpha) interactions form a 13. Han X, Sterling H, Chen Y, Saginario C, Brown EJ, et al. (2000) CD47, a ligand barrier for antibody-mediated tumor cell destruction. Proc Natl Acad Sci U S A for the macrophage fusion receptor, participates in macrophage multinucleation. 108: 18342–18347. J Biol Chem 275: 37984–37992. 33. Valk PJ, Verhaak RG, Beijen MA, Erpelinck CA, Barjesteh van Waalwijk van 14. Saginario C, Sterling H, Beckers C, Kobayashi R, Solimena M, et al. (1998) Doorn-Khosrovani S, et al. (2004) Prognostically useful gene-expression profiles MFR, a putative receptor mediating the fusion of macrophages. Mol Cell Biol in acute myeloid leukemia. N Engl J Med 350: 1617–1628. 18: 6213–6223. 34. Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, et al. (1985) 15. Jiang P, Lagenaur CF, Narayanan V (1999) Integrin-associated protein is a Proposed revised criteria for the classification of acute myeloid leukemia. A ligand for the P84 neural adhesion molecule. J Biol Chem 274: 559–562. report of the French-American-British Cooperative Group. Ann Intern Med 16. Brown EJ, Frazier WA (2001) Integrin-associated protein (CD47) and its ligands. 103: 620–625. Trends Cell Biol 11: 130–135. 35. Kaspers GJ, Veerman AJ, Pieters R, Broekema GJ, Huismans DR, et al. (1994) 17. Barclay AN, Brown MH (2006) The SIRP family of receptors and immune Mononuclear cells contaminating acute lymphoblastic leukaemic samples tested regulation. Nat Rev Immunol 6: 457–464. for cellular drug resistance using the methyl-thiazol-tetrazolium assay. 18. Tsai RK, Discher DE (2008) Inhibition of ‘‘self’’ engulfment through Br J Cancer 70: 1047–1052. deactivation of myosin-II at the phagocytic synapse between human cells. 36. Roos D, Voetman AA (1986) Preparation and cryopreservation of cytoplasts J Cell Biol 180: 989–1003. from human phagocytes. Methods Enzymol 132: 250–257. 19. Chao MP, Majeti R, Weissman IL (2012) Programmed cell removal: a new 37. Heemskerk MH, Hooijberg E, Ruizendaal JJ, van der Weide MM, Kueter E, et obstacle in the road to developing cancer. Nat Rev Cancer 12: 58–67. al. (1999) Enrichment of an antigen-specific T cell response by retrovirally 20. Majeti R, Chao MP, Alizadeh AA, Pang WW, Jaiswal S, et al. (2009) CD47 is an transduced human dendritic cells. Cell Immunol 195: 10–17. adverse prognostic factor and therapeutic antibody target on human acute 38. Kinsella TM, Nolan GP (1996) Episomal vectors rapidly and stably produce myeloid leukemia stem cells. Cell 138: 286–299. high-titer recombinant retrovirus. Hum Gene Ther 7: 1405–1413.

PLOS ONE | www.plosone.org 12 January 2013 | Volume 8 | Issue 1 | e52143 Role of SIRPa in Acute Myeloid Leukemia

39. Kaspers GJ, Zwaan CM, Pieters R, Veerman AJ (1999) Cellular drug resistance ETO-positive acute myeloid leukemia cells. J Pharmacol Exp Ther 321: 953– in childhood acute myeloid leukemia. A mini-review with emphasis on cell 960. culture assays. Adv Exp Med Biol 457: 415–421. 52. Gottlicher M, Minucci S, Zhu P, Kramer OH, Schimpf A, et al. (2001) Valproic 40. Chou TC, Talalay P (1984) Quantitative analysis of dose-effect relationships: the acid defines a novel class of HDAC inhibitors inducing differentiation of combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22: transformed cells. Embo J 20: 6969–6978. 27–55. 53. Kircher B, Schumacher P, Petzer A, Hoflehner E, Haun M, et al. (2009) Anti- 41. Hubeek I, Comijn EM, Van der Wilt CL, Merriman RL, Padron JM, et al. leukemic activity of valproic acid and imatinib mesylate on human Ph+ ALL and (2008) CI-994 (N-acetyl-dinaline) in combination with conventional anti-cancer CML cells in vitro. Eur J Haematol 83: 48–56. agents is effective against acute myeloid leukemia in vitro and in vivo. Oncol 54. Alblas J, Honing H, de Lavalette CR, Brown MH, Dijkstra CD, et al. (2005) Rep 19: 1517–1523. Signal regulatory protein alpha ligation induces macrophage nitric oxide 42. Balgobind BV, Van den Heuvel-Eibrink MM, De Menezes RX, Reinhardt D, production through JAK/STAT- and phosphatidylinositol 3-kinase/Rac1/ Hollink IH, et al. (2011) Evaluation of gene expression signatures predictive of NAPDH oxidase/H2O2-dependent pathways. Mol Cell Biol 25: 7181–7192. cytogenetic and molecular subtypes of pediatric acute myeloid leukemia. 55. Larizza L, Magnani I, Beghini A (2005) The Kasumi-1 cell line: a t(8;21)-kit Haematologica 96: 221–230. mutant model for acute myeloid leukemia. Leuk Lymphoma 46: 247–255. 43. Metzeler KH, Hummel M, Bloomfield CD, Spiekermann K, Braess J, et al. 56. Galbaugh T, Feeney YB, Clevenger CV (2010) Prolactin receptor-integrin cross- (2008) An 86-probe-set gene-expression signature predicts survival in cytoge- talk mediated by SIRPalpha in breast cancer cells. Mol Cancer Res 8: 1413– netically normal acute myeloid leukemia. Blood 112: 4193–4201. 1424. 44. Wouters BJ, Lowenberg B, Erpelinck-Verschueren CA, van Putten WL, Valk PJ, 57. Carter BZ, Kornblau SM, Tsao T, Wang RY, Schober WD, et al. (2003) et al. (2009) Double CEBPA mutations, but not single CEBPA mutations, define Caspase-independent cell death in AML: caspase inhibition in vitro with pan- a subgroup of acute myeloid leukemia with a distinctive gene expression profile caspase inhibitors or in vivo by XIAP or Survivin does not affect cell survival or that is uniquely associated with a favorable outcome. Blood 113: 3088–3091. prognosis. Blood 102: 4179–4186. 45. Nagahara M, Mimori K, Kataoka A, Ishii H, Tanaka F, et al. (2011) Correlated 58. Chao MP, Weissman IL, Majeti R (2012) The CD47-SIRPalpha pathway in expression of CD47 and SIRPA in bone marrow and in peripheral blood cancer immune evasion and potential therapeutic implications. Curr Opin predicts recurrence in breast cancer patients. Clin Cancer Res 16: 4625–4635. Immunol 24: 1–8. 46. Bullinger L, Dohner K, Bair E, Frohling S, Schlenk RF, et al. (2004) Use of 59. Isenberg JS, Maxhimer JB, Hyodo F, Pendrak ML, Ridnour LA, et al. (2008) gene-expression profiling to identify prognostic subclasses in adult acute myeloid Thrombospondin-1 and CD47 limit cell and tissue survival of radiation injury. leukemia. N Engl J Med 350: 1605–1616. Am J Pathol 173: 1100–1112. 47. Warrell RP, Jr., de The H, Wang ZY, Degos L (1993) Acute promyelocytic 60. Maxhimer JB, Soto-Pantoja DR, Ridnour LA, Shih HB, Degraff WG, et al. leukemia. N Engl J Med 329: 177–189. (2009) Radioprotection in normal tissue and delayed tumor growth by blockade 48. Chomienne C, Ballerini P, Balitrand N, Daniel MT, Fenaux P, et al. (1990) All- of CD47 signaling. Sci Transl Med 1: 3ra7. trans retinoic acid in acute promyelocytic leukemias. II. In vitro studies: 61. Soto-Pantoja DR, Miller TW, Pendrak ML, Degraff WG, Sullivan C, et al. structure-function relationship. Blood 76: 1710–1717. (2012) CD47 deficiency confers cell and tissue radioprotection by activation of 49. van Beek EM, Cochrane F, Barclay AN, van den Berg TK (2005) Signal autophagy. Autophagy 8. regulatory proteins in the immune system. J Immunol 175: 7781–7787. 62. de Vries HE, Hendriks JJ, Honing H, De Lavalette CR, van der Pol SM, et al. 50. Gozzini A, Santini V (2005) Butyrates and decitabine cooperate to induce (2002) Signal-regulatory protein alpha-CD47 interactions are required for the histone acetylation and granulocytic maturation of t(8;21) acute myeloid transmigration of monocytes across cerebral endothelium. J Immunol 168: leukemia blasts. Ann Hematol 84 Suppl 1: 54–60. 5832–5839. 51. Liu S, Klisovic RB, Vukosavljevic T, Yu J, Paschka P, et al. (2007) Targeting 63. Olek A, Oswald J, Walter J (1996) A modified and improved method for AML1/ETO-histone deacetylase repressor complex: a novel mechanism for bisulphite based cytosine methylation analysis. Nucleic Acids Res 24: 5064– valproic acid-mediated gene expression and cellular differentiation in AML1/ 5066.

PLOS ONE | www.plosone.org 13 January 2013 | Volume 8 | Issue 1 | e52143